ComfyUI Node Runs on cloud
Audio Spectrogram
Draw a clip's spectrum over time as an IMAGE, high frequencies at the top, to see hum, noise, cut-offs and harmonics, or to feed the sound to an image model.
Audio Spectrogram
- audio
- image
◄width1024►
◄height512►
◄scalelog►
◄fft_size2048►
◄floor_db-90►
◄coloursmagma►
CategoryWAS Suite/Audio
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| audio | AUDIO | The clip to draw. Every clip of a batch gets a picture. | |
| width | INT | 102464–8192 | Picture width in pixels, such as `1024`; time runs left to right. |
| height | INT | 51264–4096 | Picture height in pixels, such as `512`; frequency runs bottom to top. |
| scale | COMBO | log | `log` = octaves evenly spaced, music reads best; `mel` = hearing's own scale, speech reads best; `linear` = hertz evenly spaced, hum stands out. |
| fft_size | COMBO | 2048 | Window in samples: 4096 = finer pitch, blurred timing; 512 = sharp timing, coarse pitch; 2048 = between. |
| floor_db | FLOAT | -90-160–-10 | Level drawn darkest, in dB below a full-scale sine. -90 = quiet detail; -60 = only the loud parts. |
| colours | COMBO | magma | Colour map: `magma` and `inferno` run black to pale yellow, `viridis` purple to yellow, `greyscale` black to white. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | One spectrogram picture per clip. |